Bull Eng Geol Environ (2005) 64: 95–109 DOI 10.1007/s10064-004-0272-3 ORIGINAL PAPER

R. G. Green Rock cut on the A5 at Glyn Bends, A. Brian Hawkins North , UK

Abstract This paper discusses the dans sa traverse´e des Glyn Received: 9 August 2004 Accepted: 21 October 2004 challenges involved in the upgrading Bends, une re´gion naturelle de Published online: 31 March 2005 of the A5 to Holyhead road grande beaute´, limitrophe du Parc Springer-Verlag 2005 through Glyn Bends, an area of national de Snowdon, dans le Nord natural beauty adjacent to the du Pays de Galles. Pour le fran- National Park in North chissement d’une gorge dans des Wales. At the location where it me´tasiltstones de l’Ordovocien, une passed through a steep gorge in the partie de la route prenait appui sur Ordovician metasiltstones, part of des murs de soute` nement de 10 m de the original road was supported on hauteur et plus de 200 ans d’aˆ ge. Ces 200 year old, 10 m high retaining me´tasiltstones contiennent quelques walls. The metasiltstones contain niveaux de cendres volcaniques de some low strength bands of volcanic faible re´sistance me´canique, ayant ash, on which a planar failure had rendu possible un processus de rup- taken place prior to the construction ture plane avant la construction de of the original road. This paper dis- la premie` re route. L’article pre´sente cusses the up-grading of this part of comment cette section de l’A5 a e´te´ the A5, which involved a ame´liore´e, avec la re´alisation d’un 550 m long, 25 m deep cutting and de´blai de 550 m de long et 25 m de an 18 m high embankment. This hauteur et d’un remblai de 18 m de work was designed to minimise the hauteur. Les travaux ont e´te´con- visual impact in this environmentally duits afin de minimiser l’impact vi- sensitive area, while limiting the suel sur cette re´gion naturelle, tout R. G. Green interruption to traffic on a signifi- en limitant au mieux l’interruption Parsons Brinckerhoff, Calyx House, cant trunk road. de trafic sur cette route nationale. South Road, Taunton, TA1 3 DU, UK E-mail: [email protected] Mots cle´s De´blai rocheux Æ Fax: +44-1823-424401 Keywords Rock cut Æ Blasting Æ Pre-split Æ Roads Æ Abattage a` l’explosif Æ A. B. Hawkins (&) Pre´-de´coupage Æ Routes Æ Nord du Charlotte House, 22 Charlotte Street, Bristol, BS1 5PZ, UK Re´sume´L’article pre´sente les de´fis a` Pays de Galles E-mail: [email protected] relever pour la modernisation de la Fax: +44-117-9273994 liaison routie` re A5 entre Londres et

Introduction of Holyhead. In view of the mountainous topography in North Wales, the road followed the coast, through In the 1700s, the route from London (Eng- . With the increasing traffic in the early 1800s, land) to () was via the North Wales port ways of shortening this arduous journey were considered 96 and Thomas , subsequently known as one of Britain’s great civil engineers, was commissioned to examine the feasibility of creating a more direct route through , and Betws-y-Coed (Fig. 1). After overcoming some difficult engineering problems, 135 km of road between Shrewsbury and Bangor was opened in 1819 and continued as the main A5 trunk road until the 1990s. Between and Betws-y-Coed, a 2.3 km stretch of the A5 is particularly twisty as the road passes through the gorge of the Afon Ceirw. Telford’s design involved the cutting of a ledge into the rocks along the northern side and, where necessary in the hollows, constructing high retaining walls (Fig. 2). While this Fig. 2 Part of the 200 year old retaining walls supporting the 6 m wide road was acceptable for stagecoach travel, the southern side of the original A5. Also seen is the lower part of a sharp bends, with radii of only 30 m, did not meet rock buttress present on the northern side (Fig. 3) modern requirements. The road had no verges and there was frequently a sheer rock slope on the northern edge (Fig. 3) and a 30 m drop into the river gorge to the south. In addition, with over 50% of the route between Ty-nant and Dinmael (Fig. 4) having a visibility of less than 70 m, the safe road speed was often no more than 20–40 kph. In 1989, it was decided to make the North Wales coast road between Chester and Bangor (A55) the main route to the Holyhead–Dublin ferry and, in view of the engineering problems and the environmental sensitivity of the area, to make only selected improvements to the A5 to bring it up to modern road standards. On inspecting the highway at walk-over stage, there was considerable concern that a wedge failure could occur in the rock buttress 6–12 m above the road at the eastern end of the gorge section (Fig. 3). In addition, examina- tion of Telford’s 10 m high retaining walls showed that in one limited area the slope below the structure had failed and undercut the toe of the wall, leaving the masonry unsupported. Fig. 3 Sheer rock face immediately north of the A5 trunk road

Fig. 1 Location of the Glyn Bends section of the A5 London to Holyhead Road 97

Fig. 4 General contour map showing the location of the Afon Ceirw gorge through which the original Glyn Bends section of the A5 road was constructed (1 km Squares)

This paper describes the topographical and geological setting of the Glyn Bends section of the A5 and con- Topography siders the evolution of the improvements. It also dis- cusses the problems encountered during the North Wales contains the highest land in and construction, due mainly to the presence of thin volcanic Wales with peaks of over 1,000 m a.s.l., including ash bands within the strong Ordovician metasiltstones, Mount Snowdon (west of Betws-y-Coed), which rises to which in this area form a southwards plunging asym- 1,085 m. In the Glyn Bends area, the relief is lower with metric anticline. An important consideration was the Mwdwl eithin rising to 470 m and Foel Goch to 611 m, visual impact of the new road in an area of open to the north and south of the A5, respectively. These countryside, only some 10 km from the Snowdonia hills form the watershed between the rivers draining National Park (Fig. 5). westwards and northwards to the and those draining eastwards towards England. As a consequence, west of Corwen the A5 road passes up the valley of the Afon Ceirw, a tributary of the eastward-flowing River Dee, before descending towards Betws-y-Coed following the Afon Merddwr, a tributary of the northward-flowing River (Fig. 5). Although the watershed is not particularly high, the eastward-flowing rivers have eroded deep gorges as they descend from the almost flat-topped Moor through the Lower Palaeozoic metasediments. The River Ceirw meanders across the alluvial infill of a glacial valley, where the profile drops at some 1:50– 1:60 to the west and east of Glyn Bends respectively (Figs. 4 and 6). In the gorge south of Pen-y-bont, however, the river falls 20 m in 350 m (1:18). Although the Ordnance Survey maps indicate waterfalls in this area, in reality they are a series of rapids which are most notable where the river falls 10 m in a distance of only 80 m. As seen from the contours, above the gorge the ground is more gentle and open in the area of Pen-y- bont. When Telford designed his highway in the 1810s, Fig. 5 Sketch map showing the A5 and alignments and in order to maintain a realistic gradient for the main river valleys referred to in the text he placed much of the road along the northern slope 98

Fig. 6 Detailed topography north of the Afon Ceirw gorge (2 m contour spacing) showing the area where the planar fail- ure occurred and the alignment of part of the new roadway

of the gorge in 6 to 12 m high cuts and built masonry the road re-alignment scheme was out to tender; hence 7 walls where the road crossed small stream courses/hol- the modern map was not used in the initial planning and lows. In the Glyn Bends area, the road is some 30 m site investigation stage. above the river level. Between the two main bluffs, the As can be seen from Fig. 7, the bedrock geology in northern slope is more gentle, rising at circa 30 for the area is Lower Palaeozoic in age. The region was approximately 40 m (Fig. 6). Above this, there is a 4 metamorphosed during the Caledonian Orogeny (Silu- to 6 m high cliff, which in places is overhanging. As rian–Devonian) and as a consequence the individual discussed later, the relief in this area has been created by lithologies, frequently referred to in the geological lit- a rock slide, which clearly pre-dates the construction of erature by their sedimentary terms, from an engineering the road. point of view behave as metamorphic rocks. The On the flatter area north of the gorge, there are sev- 1:50,000 map (Corwen Sheet 120, Fig. 7) gives the fol- eral ridges almost at right angles to the river. It is likely lowing stratigraphic succession. that it is the presence of stronger, less fractured hence more resistant rocks which is responsible for the nar- Ashgill series Maerdy Mudstone Formation (up to rowness of the gorge and the main area of rapids dis- 1,600 m) with the Rhiwlas Limestone (0–5 m) at the cussed above. base Caradoc series Tyn-y-glyn Mudstone Formation Background geology (0–2 m) Gelli-grin Calcareous Ash Formation (0–200 m) including the Cymerig Limestone The primary mapping at the 1:63,360 scale was under- Allt Ddu Formation (90–210 m) taken by Burke, Aveline and Ramsey and published in Glyn Gower formation (160–230 m) 1885. This remote area was not mapped again for over a century, but in the meantime, Alder had undertaken his The Glyn Gower Formation consists of well-bedded, doctoral studies in the area (1976) and the local laminated/cross-laminated muddy siltstones, siltstones authority had commissioned Webb to carry out some and sandstones with spaced cleavage. Fossils are rare. At specific mapping in the Glyn Bends region. When the the top of the formation is the Frondderw Tuff forma- 1:50,000 geological map, including much of the work tion (0–14 m), which consists of four sub-marine debris previously undertaken by Alder, was published in 1993, flows with the individual bands fining upwards. 99

Fig. 7 Enlargement of the relevant part of the 1:50,000 geological map of the area (Sheet 120)

The Allt Ddu Formation includes siltstone horizons of the gorge. Although the stratigraphy of the limestones with a number of ash/tuff bands; the most prominent is difficult to distinguish, those in the west are now being the Garth Tuff. Some of these pyroclastic bands considered to be the Cymerig Limestones and those to are sufficiently thick to form mappable units in parts of the east the younger Rhiwlas Limestone, which is in- North Wales. In places, the siltstones are highly fossil- cluded within the Maerdy Mudstone Formation. These iferous. limestone horizons are particularly important, as karstic The Gelli-grin Calcareous Ash Formation is typically features have developed within them. Indeed, during the volcanoclastic material, which formed as an airfall accu- walk-over surveys, small cave systems were identified in mulate in a marine environment. As well as containing the eastern part of the original road sections. occasional fossils throughout the main rock mass, a The Tyn-y-glyn Mudstone Member is less than 2-m number of distinct fossiliferous horizons are present. At thick, but was identified by Alder as a dark pyritiferous outcrop, such as seen along the A5 road cutting, the fossils non-calcareous quartz-veined mudrock. have weathered selectively. The Gelli-grin also contains The Maerdy formation, which unconformably over- the approximately 6 m thick Cymerig Limestones, which lies the Gelli-grin, is typically a thick sequence of silt- consist of oval gravel to boulder-sized concretions stones with some sandstone horizons. As mentioned of impure limestone, sometimes in a silt-rich matrix. above, the Rhiwlas Limestone generally forms the basal Although Alder had initially considered only one beds of this formation. limestone horizon to be present in the Lower Palaeozoic The Gelli-grin dips between 30 and 65 towards the strata of this area, subsequently two bands were identi- south, although from the exposures it was noted that fied and can now be seen at the western and eastern ends there was a 60 splay of orientations between N150 and 100

N210. In many outcrops, particularly in the siltstones, being four times the national average for this type of it is difficult to distinguish bedding from stress-induced road; spaced cleavage unless fossil horizons or minor litho- (c) structural condition of the existing retaining wall logical changes are present. In addition to the dominant supporting the A5 some 30 m above the base of cleavage, which dips northwards between 35 and 75, Afon Ceirw gorge; there is a second sub-vertical set. (d) potential instability of a wedge of rock at the top of During the Quaternary, when the area was glaciated, the buttress seen in Fig. 3, part of which had already large-scale removal of the weaker lithologies would fallen. undoubtedly have taken place such that in addition to the extensive area of till in the western part of the pro- As a serious failure of the retaining wall could close posed route corridor, glacial deposits accumulated be- the road and there were no suitable diversion routes in tween some of the rock ridges on the flatter land above this highland region, the Client considered the the gorge. With meltwater release as the thick glaciers improvement of the road should be undertaken with the were receding, intense erosion would have taken place, minimum of delay. The design was to comply with the at least deepening the Afon Ceirw gorge. Guidance Notes ‘‘Roads in Upland Areas’’ (Anon 1990) While working for County Council, Webb and for a speed of between 80 and 100 kph. However, identified two areas of landslip. To the south west of the Authority emphasised the sensitivity of the natural Pen-y-bont Farm a small disturbance was located and environment, immediately adjacent to the Snowdonia 200 m to the south east of the farm, the hollow left by a National Park. larger landslip is evident both in the field and on the Initially five routes were assessed, four to the north of scheme contour map (Fig. 6). This slip, near the contact the gorge and one to the south. On environmental, between the calcareous siltstone and the Rhiwlas constructional and economic considerations, the Tyn-y- Limestone, covers approximately 1,250 m2. Upslope is a glyn route was the most favourable (Fig. 4), having 4–6 m high rock face which in places overhangs and has the least effect on the existing properties and involving no clearly retreated by toppling failure since the main mass additional trafficking of the stone masonry arched bridge movement. On close examination of the base of the rock (Pont-y-Glyr-diffwys) across the head of the gorge. faces, a very thin band of weathered ash was identified in The final horizontal alignment was chosen to the northern extremity of this wedge-shaped hollow and (a) optimise the distance from the A5 retaining walls to hence it was considered that this section the gorge slope allow blasting to take place without unduly had failed along such a low strength band. increasing the depth of the cut (Fig. 8); (b) provide buttress support to the rock-anchored Scheme design retaining wall east of Tyn-y-glyn, which had been built some years ago to support the highway; (c) obtain a cut/fill relationship with an acceptable Improvement of the road between Ty-nant and surplus, as disposal of fill would be easier and Dinmael was necessary because of the cheaper than importing additional material; (a) extremely poor alignment, with the consequential (d) provide a gradient and visibility for a design speed restrictions on the safe, free flow of trunk road of 85 kph; traffic; (e) limit the land takes and minimise the aesthetic (b) poor accident record, with the overall personal intrusion of the new road into the existing environ- injury accident rate between 1984 and 1987 ment.

Fig. 8 Schematic drawing showing the relationship be- tween the new road cut and the original A5 supported by masonry retaining walls on the slope of the gorge 101

The improvement works included a 550 m long cut As seen in Fig. 6, the land rises to the north of the between Pen-yr-erw in the west (Ch 400) and the point new alignment and from early analysis of the exposure where the new road would cross the existing A5 (Ch 950). data it was appreciated that the main geological struc- The 7.3 m carriageway would be widened to 10 m in the ture is a southward plunging anticline. As a conse- eastern approach to the cut, to provide a climbing lane. quence, it was considered essential to undertake In order to reduce the size of the cut and thus the impact drainage of the rock mass, particularly north of the of the road on the environment, this 10 m wide section road, so that water flow would not be inhibited by the of roadway had only 0.3 m hard strips and 2.15 m ver- presence of low permeability ash bands. This could lead ges, which included a rock ditch and fence. Steep slopes to a build-up of water pressures behind bands of low were a key feature of the design to reduce land-take. East strength material at a time when the lateral constraint of Tyn-y-glyn, a high embankment was designed to re- was being removed. In order to optimise their effect and duce the gradient of the road, use much of the minimise the visual impact, the permanent drains were 0.25 million m3 of rock fill material generated in the cut designed with a fan configuration, permitting one col- and provide additional buttress support to the lower part lecting point for every five drains (Fig. 11). The col- of the hill in which instability had occurred in the past. lecting areas were to be recessed and hidden behind In view of the strength and lack of discontinuities in masonry walls built of site-won rock. the metasiltstones, it was clear from the outset that any significant rock excavation would require blasting. It is well known (e.g. Hoek and Bray 1971; Matheson 1986) Site investigation that unless pre-splitting is used the bulk blasting required to loosen and disintegrate the rock in the main excava- When the authors became involved in the scheme, a site tion will disturb the strata behind the designed face. A investigation had already been undertaken under the decision was therefore taken to use a pre-split cut of 3:1 supervision of Clwyd County Council. In addition to the (72) in the main cut. Above this, where overburden and intrusive investigation, discontinuity mapping of the weathered rock is present over part at least of the cut exposures was undertaken and an interpretative report length, angles of 1:2 (26) were planned (Fig. 9). How- prepared. In view of the failure to the north of the ever, the design guide ‘‘Roads in Upland Areas’’, pro- existing road (Fig. 6), this earlier report had concluded duced by the Welsh Office required that the highway that side slopes as low as 1:2 (26) would be required, should have as natural an appearance as possible. Con- necessitating a very wide cutting in this hilly topography. sequently, rather than have a consistent straight/curved Following the re-appraisal, a supplementary site inves- 3:1 pre-split face, the design included ledges in the 3:1 slope and some batters as steep as 4:1 (76, Fig. 10). An important consideration was the impact of the construction works on the existing major trunk road. While it was desirable to minimise interruptions to traffic flow, it was clearly important to protect road users from the shock of an explosion and, following each blast, to ensure both that the retaining walls were not affected and no fly-rock had breaked the protection measure and landed on the road. As a consequence, blasting was generally permitted only once per day when the traffic flows were halted for some 20 min.

Fig. 10 Designed variations to the standard 3:1 batter to provide Fig. 9 Designed batter angles through the new rock cut at Glyn irregularity to the cut face (‘‘monkey ledges’’), thus reducing the Bends visual impact 102 tigation was considered necessary. The purpose of this (e) the ground water regime and its significance for the investigation was to establish short- and long-term stability of the rock and soil slopes; (a) whether a 20–30 m cut could be safely constructed in (f) the nature of the material to the east of the cut on this geological setting; which the embankment would be constructed, in (b) if the rock blasted in the cut would form satisfactory order to determine a safe batter angle; rock fill; (g) the thickness of Telford’s retaining walls to the (c) the nature of the glaciogenic deposits believed to south of the existing A5 and the nature of the fill exist at the western end of the cut and assess the between the walls and the bedrock. maximum stable slope angle; (d) whether the glaciogenic material excavated would After a detailed walk-over survey, during which all form satisfactory earth fill or could only be used for exposures were re-examined, a site investigation was landscaping; designed which involved 29 trial pits (sometimes ex- Fig. 11 Details of the fan drains; plan section, collecting point and recessing 103 tended into trenches) and 15 rotary cored boreholes. Although the cutting was between 20 and 30 m deep, trial pits were considered necessary to establish the reason for the rock ridges referred to above and to in- spect the nature of the near surface weathered ash bands such as that contributing to the historic planar failure indicated in Fig. 6. In order to minimise fracturing of the rock due to the torque of the drill bit, 75 and 92 mm cores were obtained. It was considered essential for both the design engineers and the contractors to be able to appreciate the massivity or broken nature of the ground and hence, in addition to the total core recovery (TCR), solid core recovery (SCR), rock quality designation (RQD), fracture frequency (FF) and maximum core length (MCL), a pictorial profile of the discontinuities was included on the log sheets. Below Fig. 12 Shear strength failure envelope obtained for a discontinuity the overburden level, the RQDs were generally in excess in the metasiltstones at 16.17 m in BH 66 using the Hoek shear box of 70% and at many levels there were 0–2 fractures per metre. For practical purposes, therefore, the visual ment of the ground conditions. For this to be efficient, fracture log was of considerable value in distinguishing the site team included a full-time engineering geologist the intact from the more fractured zones. to undertake logging of the faces and a Deputy Resident Initially, it had been intended to supplement the core Engineer with a geotechnical background. In addition, logs with a visual inspection of the holes using CCTV, the writers would visit the site regularly and at any time but the downhole conditions were not ideal. It was then when features of concern became apparent or difficulties decided to use a slim line geophysical logging technique, were encountered. the dipmeter survey, in which the sonde is capable of With regard to the masonry walls, it was not possible measuring a number of formation characteristics to find any specific details of these structures, which were including the radioactivity, the inclination of the strata up to 10 m high. However, contemporaneous literature and the orientation of the maximum dip, as well as the describing walls in the region suggested that they would integrity (caliper log) and deviation of the hole. In view be set on a rock ledge, with a wide base and narrowing of the importance of locating weak horizons, which profile (Fig. 14). could be lost by drill flush, the caliper logging was The site investigation for the walls included sub- particularly valuable. horizontal drilling through the masonry and backfill into In order to establish some shear strength values, the bedrock behind and vertical cored holes through the samples were tested in the Hoek shear box. Figure 12 road. The data from the horizontal holes showed the shows a typical result from a test carried out on a dip- walls were thinner than had been anticipated, varying ping discontinuity in the metasiltstones at 16.17 m in BH between 1 and 1.4 m. Some holes indicated a reduction B66 which gave a /value of 47. In the same borehole, a in thickness with depth, suggesting that the lower part of 5 mm thick soft greenish grey clay with a greasy feel was recovered at 21.3 m. When a sample including this moist clay was tested in the Hoek shear box it gave a / of only 14 (Fig. 13). The trial pitting exercise along the line of the ridges confirmed the presence of faulting. At the top of the planar failure (Fig. 6), a trial trench exposed the low strength ash band on which the historic mass movement probably took place. Also seen were two discontinuities on which moist powdered metasiltstone closely resem- bled the ash band. From the trial pitting exercise it was clear that the number of thin volcanic ash bands was likely to be more than even high quality coring would recover in strong inclined rocks as the weak material could easily be lost in the drill flush. As a consequence, it was decided that the contract would need to allow for changes in the design Fig. 13 Shear strength failure envelope obtained for the moist ash as the work proceeded, based on a continual re-assess- band material at 21.3 m in BH 66 using the Hoek shear box 104 the wall provided only a facing to the original slope Construction while the upper part was used to retain the fill material. In some cases it was not easy to distinguish the It was appreciated from the outset that the upper surface superficial material from the in situ weathered bedrock of the main rock cut would vary significantly. The ridges in the horizontal holes. In the upper holes, the material running at approximately right angles to the river would was invariably a red brown gravel with much clay, require blasting from the existing ground surface while which it was considered had been imported for the the glaciogenic materials in the hollows between the construction of the road. At lower levels, the recovered ridges could easily be dug by backactors. Invariably, the material consisted of more blocky gravel, cobble and rocks were more shattered (Figs. 15 and 16) adjacent to boulder-sized metamudrocks with some red brown hollows where faults existed and hence again could be clayey seams; the open joints indicating this was easily dug using a machine with a rock bucket. In view probably in situ weathered material. The bedrock of this anticipated variability, the upper part of the rock recovered by the coring included some light grey cut was designed to have an angle of 1:2, which in good tuffaceous/ash material. material could steepen to 1:1.5 (Fig. 9). In the main rock The horizontal and vertical drill holes indicated an cut, however, blasting would be required throughout irregular rock head profile, probably reflecting the and steep pre-split slopes would be formed as discussed change in topography seen elsewhere in the gorge. Some above. evidence of voiding was recorded during the drilling and The pre-split for the main cut was drilled and fired in on the CCTV survey. In addition the driller’s record 9 m long panels. The drillers were specifically requested showed cavities of 50–100 mm and in two instances a to record where the rocks were more shattered and steel tape was inserted into the hole between drill runs whether the advancing hole passed through low strength for a distance of 1.5 m more than the drilled depth. As volcanic ash bands. In addition, the advance of the pre- seen on the subsequent CCTV film, the cavities were split was restricted to 9 m ahead of the bulk blasting in narrow, linear features, probably representing open order that the cut face could be inspected and a decision joints/fissures associated with the stress release of near made as to whether the 1 m spacing of the pre-split holes surface rocks at the side of the gorge. should be reduced and/or the charge modified. Failure of these retaining walls during the construc- Brief details of the blasting parameters are given in tion works would have created a major traffic problem in Table 1. The vibration constraints stipulated during the this part of highland Wales where the only alternative design phase proved to be onerous, particularly where routes involve long detours. To reduce the impact of the blasting was undertaken within 50 m of the walls. necessary blasting, which could have produced signifi- Indeed, in these areas, it was necessary to reduce the cant distress to the retaining walls, the peak particle number of holes detonated, particularly in the pre-split velocity was limited to 10 mm/s. In addition, a signifi- panels, in order not to exceed the required ppv value. cant sum of money was set aside for monitoring during Although this often resulted in a reduction of the panel the works, including the provision of scaffolding to al- size, it is considered this approach was justified as the low ready access to inspect the masonry. integrity of the walls was not compromised. As there was some concern that the permeability of the rock may not allow effective de-watering using ver-

Fig. 14 Sketch of a typical wall construction as recorded in the literature in the early 1800s Fig. 15 Early cut showing variations in the massivity of the rock 105

Fig. 16 Rock face to the west of Fig. 15 showing one of the ridges referred to in the text Fig. 17 Schematic diagram showing excavation phases tical pumped holes, the excavation involved a ‘‘canal’’ cut (Fig. 17), which facilitated the installation of into the cut at 30 was exposed and a planar slide took sub-horizontal drain holes, extending well behind the place (Fig. 19). Despite this failure, the cutting could designed cut face. Installing these drain holes at right still be accommodated within the existing land-take by angles to the rock cut would ensure that there was no omitting the narrow drainage berm at the top of the build-up of water behind the face prior to the con- main rock cut, modifying the berm drainage design and struction of the permanent fan drains (Fig. 11) while undertaking some re-profiling. In addition, the over- maintaining the significant lateral burden necessary for burden was netted to prevent coarse material moving an effective pre-split. The canal cut also had the across the smooth failure surface and ricocheting onto advantage of allowing an inspection of the geology some the road below. 9 m in front of the designed northern face. This slip was high in the rock face and involved only The contract specified that the excavation lifts should some 200 t of material. However, as the excavation not exceed 6 m. This would allow visual assessment of progressed, the anticipated low strength horizons were the rock face and the installation of the designed sta- exposed (Fig. 20) at higher angles and at varying ori- bilisation and any additional remedial measures, with- entations to the cut face (Fig. 21). A decision was out having to use large hydraulic lifts as machines could therefore taken to install 194 mm steel tubes set into operate from ramps of fill. In the area of the axis of the 250 mm diameter holes with a cementitious grout plunging anticline, however, (Fig. 18), there was concern (Fig. 22). These were subsequently increased to 300 mm that the north face may fail prior to the anchors being as drilling tolerance was difficult to maintain in the in- installed and stressed. Over this 150 m length, in the top clined strata consisting of strong metasiltstones with part of the cut the mucking-out was reduced to 3 m lifts weak horizons. These would act as dowels to retain the and stabilisation measures were undertaken before fur- ther deepening. In one location, however, some over- deepening did occur. A low strength horizon dipping

Table 1 Details of the blasting at Glyn Bends

Pre-split Bulk

Purpose To form stable Necessary for excavation rock faces of 0.25 million m3 Drill holes Inclined (72) Vertical Hole diameter 105 mm 105 mm Hole spacing 0.85–1.0 m 2.5·2.5-m grid Blast ratio 0.25–0.35 kg/m2 Fragmentation ratio ca. 0.45 kg/m3 Vibration constraint 10 mm/s (pv) 10 mm/s (ppv) No. of holes per 3–10 Double and triple deck Fig. 18 Photograph of the south face showing the southerly delay charges plunging anticline 106

Fig. 19 View of the area where the planar slide took place in the Fig. 21 A further example of areas where slides took place, in part upper part of the north cutting related to the increasing dip on the limb of the anticline slope until the appropriate long-term stabilisation mea- restraining plate could be fixed at berm level (Fig. 22). sures could be installed. Along part of the 150 m of cut, The installation of this form of dowel generally proved two rows of 10 to 12 m long dowels of 12.5 mm wall very successful in supporting the rocks in the main cut thickness were installed in the 6 m wide berm at 1.5 m beneath the berm level during the subsequent excavation centres, off-set at 1.5 and 3.5 m from the cut line. In phase, prior to the installation of the permanent anchors other areas, this was reduced to one row at approxi- (Fig. 23). mately 4 m centres. The decisions were not based solely The working loads for the mono bar anchors were on mathematical analyses but were very much an engi- determined as 400 and 600 kN; their generalised eleva- neering judgement in this specific site situation where the tions and inclinations are shown in Fig. 23. In the case rock face was to be excavated by pre-split and bulk of the upper batter, the fixed lengths were placed behind blasting. In addition to ensuring the short-term interg- a line projected at 1:2 from the back of the intermediate rity of the works, there was a need to consider the safety berm. The upper row of anchors in the main cut was also of personnel working below the face. placed behind a 1:2 line projected from the rock catch It was appreciated that the firing of the pre-split and drain. The middle and lower anchors, however, had bulk blasting in the main cut would damage the cement fixed lengths at least 8 m behind the cut face so as not to grout around the dowels. To restrain the dilation/uplift over-stress the rocks in the slope. A total of some of the rocks when the blasting took place, a rock bolt 4,500 m of anchors was installed in the cutting, inclined was installed within the dowel tubes such that a at between 10 and 15 to the horizontal. The geological mapping of the face and interpretation was ongoing throughout the construction period (Fig. 24). Most of the strata were strong to very strong metasiltstones although some limestones occurred; the boundary between the two lithologies being gradational. The volcanic ash bands exposed varied in thickness be- tween 50 mm and 1.3 m. In some of the thicker bands, there were pipe-like sedimentary flame structures and in some cases, crystal growth slickensides were pres- ent—evidence of slow tectonic movement in a wet, chemically saturated environment. As the cutting was developed, clear evidence of an open fold was observed (Fig. 18). The general struc- ture of the area between Ch 650 and 950 is shown in Fig. 25, where it can be seen that the crest of the anticline occurs at approximately Ch 820 and the hinge of the adjacent syncline at approximately Ch Fig. 20 A similar situation to Fig. 19, but to the east where the 740. The folds plunged in a southerly direction be- angle of inclination is steeper; some bolts have been emplaced and a tween 20 and 30. As a consequence, to the east and wall is being constructed west of the main anticline and syncline the bedding 107

Fig. 22 Details of the 10 to 12 m long dowels and plates to inhibit lateral and dilation/up- lift movement

was generally perpendicular to the design face effectively parallel to the cut face. The second set were (Fig. 16). However, between Ch 780 and Ch 900, the also high angled but with a north–south orientation. As strike gradually swings round to a more easterly such they were generally perpendicular to the design face direction (Figs. 20 and 21) making the face more and had little effect on the stability. susceptible to wedge-type failures when intercepted by As discussed above, in order to minimise the visual joints or faults. impact, ‘‘monkey ledges’’ and steepened excavations The faults crossing the area can be separated into two were to be cut at the positions shown on the drawings, sets. Strike/slip faults orientated in an east-west direc- unless on site inspection indicated that these positions tion are near-vertical and have crystal growth slicken- were inappropriate. In view of the considerable struc- sides up to 30 mm in thickness. These east–west tural disturbance of the rocks in which the pre-split took structures often occurred at a short distance behind and place, however, the need to create irregularities was limited. On-going monitoring was undertaken on the high retaining walls on the south (gorge) side of the existing A5 road using scaffolding erected along their entire length. After each blast, the walls were checked to ensure

Fig. 23 Schematic cross section of the northern slope showing the rock anchors and dowels installed in the middle part of the cut Fig. 24 Face mapping using hydraulic lift 108 the integrity of the road prior to the highway being re- opened to traffic. This paper has concentrated on the northern side of the cutting as this is where the main problems were anticipated in view of the southerly dipping anticline. The berm on the southern side was narrower and less remedial work was undertaken as the inclination of the strata is into the face. However, as noted above, some of the discontinuities, both joints and faults, run effectively north west/south east (parallel to the road in one area) and hence bolting was required. A typical example of such a joint is shown in Fig. 26. Figure 27 shows the northern slope of the com- pleted road through the rock cut. At the western end, where the cutting was through glaciogenic deposits, the Fig. 26 Steeply dipping planar joint cutting the southern slope with designed slope of 1:1.75 was satisfactorily created with an orientation almost parallel to the road no major difficulties. Slope drains were installed as springs were known to exist, emanating from the buried rock surface. The embankment to the east of the rock cutting was constructed mainly using the 0.25 million m3 of site-won rock fill and covered with a veneer of glaciogenic material, generally at an angle of 1:3.

Summary

The new rock cut at Glyn Bends on the A5 in North Wales was undertaken to alleviate the difficult condi- tions where this major trunk road passes through a gorge up to 40 m deep. When the road was constructed some 200 years ago, rock cuts were undertaken to create

Fig. 27 View of the completed road showing the finished northern face with some ‘‘monkey ledges’’ cut to break the line of the pre- Fig. 25 Summary of the structural geology along the northern pre- split slope split slope of the rock cut 109 ledges along part of the gorge while in other areas the view of the planned stabilisation measures and the 6 m wide road was supported by up to 10 m high likelihood that some additional work would be required, retaining walls. the excavation lifts were restricted to 6 m (to provide The Glyn Bends area, between Corwen and Betws-y- access) and 3 m in the sensitive zone near the axis of the Coed, is close to the Snowdonia National Park—an area anticline. Despite this, a small amount of overdig caused of outstanding natural beauty. As a consequence, the slope movement, emphasising the significance of con- client and designers were very concerned to create a re- trolling the depth of dig and the timely emplacement of alignment which was in sympathy with this upland anchors and drains. environment. Although the minor slip in the upper cut face was The strata through which the main 550 m long, accommodated without the need for further land-take, 25 m deep rock cut was created were mainly Ordovician. some additional strengthening of the lower slope was These rocks were metamorphosed during the Caledo- considered necessary before the cut was extended below nian Orogeny hence the main lithologies in which the cut the main berm. It was therefore decided to install a series was undertaken were metasiltstones with two limestone of 10 to 12 m long dowels through the northern mid- horizons. In addition, some volcanic ash bands within slope berm to reduce the possibility of rock sliding when the Ordovician strata formed weaker horizons, in which the final cut was undertaken. By placing a bolted anchor crystal growth slickensides had sometimes developed. into the dowels, it was also possible to restrict uplift/ Near the surface these ash bands had often weathered to dilation which could cause the rocks beneath the berm to form horizons with low shear strengths. become loosened during the blasting due to lack of In this area, the Caledonian Orogeny imposed an vertical load. These dowels also made a positive con- open fold structure, with an anticline and syncline ex- tribution to the overall stability of the rock slope. posed in the cut. The folds plunge southwards such that A vibration constraint of 10 mm/s was imposed and the strata dip out of the high northern face. this, coupled with ongoing inspection, ensured that the In view of these difficult ground conditions, it was integrity of the adjacent 200 year old retaining walls was particularly important that the exposed geology was not compromised. mapped and ongoing assessments made. For this reason, In order to reduce the visual impact of the cutting the the Deputy Resident Engineer had a geotechnical design included variations on the cut angle and the background and an engineering geologist was on site creation of irregular ‘‘monkey ledges’’, particularly on full-time during the excavation of the rock cut. the higher northern slope. The design took account of the fact that adjacent to This paper discusses the investigation and design the high rock cut the existing A5 was supported by an philosophy and highlights some of the constructional anchored wall over an area where movement had oc- activities undertaken to create a significant cut through curred in the past. To protect the longevity of this old, at Glyn Bends North Wales. already anchored wall and to produce an acceptable grade to the new road, a high embankment was built Acknowledgements The authors wish to thank the Director of from the 0.25 million m3 of rock taken from the Highways, National Assembly for Wales, for permission to publish this paper. Any of the statements or comments made above should cutting. be regarded as personal and not necessarily those of the National The design involved 4,500 m of anchors as well as Assembly for wales and constituent part or connected body. extensive drainage of the steep 3:1 pre-split cut slopes. In

References

Alder K E (1976) The geology of an area Anon (1990) Roads in upland areas—A Matheson GD (1986) Design and excava- around Bettws Gwerfil Goch and Me- design guide. Highways Directorate, tion of stable slopes in hard rock, with lin-y-Wig, Clwyd. Unpublished PhD Welsh Office, Cardiff particular reference to pre-split blasting. Thesis, Clare College, Cambridge Hoek E, Bray JW (1971) Rock slope engi- In: Conference on Rock Engineering neering. Inst Mining Metall Lond and excavation in an Urban Environ- ment. Inst Mining Metall Lond, pp 271–283